Generating Reports and Web Apps

Size: px

Start display at page:

Download "Generating Reports and Web Apps"

Joseph Whitehead
5 years ago
Views:

1 Generating Reports and Web Apps Module 10

2 Today s Agenda Installing software to use Markdown on your own machine Walkthrough of Markdown and markup languages more generally A bit of a preview of next week In-depth on shiny How to create your own shiny apps How to enable other people to access your shiny apps

3 Necessary Installs Beyond R R Packages (not actually automatic) rmarkdown shiny knitr ggvis Software for Exports PDF: LaTeX ( get MiKTeX [Win] or MacTeX [Mac]) and set it to automatically install packages during installation HTML: pandoc v ( for now) DOCX: MS Word Although you could theoretically use RMarkdown on the command line/directly to generate reports, we will only use R Studio. 3

4 R Markdown utilizes a Markup Language Markup languages Plain text document annotation systems intended to provide text meta-data directly on a manuscript The term "markup" comes from traditional marking up by hand of documents Examples of markup languages HTML LaTeX Markdown Advantages vs. disadvantages (comparing WYSIWYG vs. WYSIWYM) Utilize plain text, so they are universally accessible and will not become obsolete Utilize plain text, so they are hard to read and require computer interpretation 4

5 Example of HTML Markup 5

6 Example of Markdown Markup (from Wikipedia) 6

7 Markdown Signals See Help/Markdown Quick Reference in R Studio Or use this cheat sheet: /wpcontent/uploads/2016/0 3/rmarkdowncheatsheet-2.0.pdf Or use this official guide: wpcontent/uploads/2015/0 3/rmarkdownreference.pdf 7

8 Embedding Code `r codehere` to execute that code wherever you want to execute it. To execute code as a block (and display it as such): ```{r} codehere ``` Options for the {r} call, separated by commas (also see cheat sheet): name: name the code echo=true: display the code (does not affect results or messages) message=true, warning=true, error=true: display messages, warnings, and errors eval=true: run the code results='markup': other options: "asis", "hold", "hide" Full list here: Example ```{r mycode, echo=f} myvar <- 1 ``` 8

9 Basic YAML Headers Same format every time --- title: "My Title" author: "My Name" date: "Today" output: output_type --- output_type specifies what you're going to generate html_document pdf_document word_document beamer_presentation # PDF slideshow slidy_presentation # HTML slideshow using slidy template ioslides_presentation # HTML slideshow using ioslides template md_document # a.rmd markdown file so when is this useful? 9

10 Extended YAML Options Indention and tabs matter, which do not follow clear patterns except re:output options Output options vary by output format, but HTML is by far the most flexible Again, see title: "My Title" author: "My Name" date: "Today" output: html_document: theme: cosmo number_sections: true toc: true toc_float: true --- Most flexibility with display comes from creating CSS files 10

11 CSS: Cascading Style Sheets HTML is a markup language intended to provide webpage content CSS is a style language intended to provide stylistic/formatting data for markup HTML+CSS is the backbone of the web We will talk about CSS more later when we get into web scraping but will focus more on how-to-read than how-to-write If you want to learn to write CSS, I suggest Codecademy 11

12 HTML + CSS <html> <head> <title>my Webpage</title> <link href="mycss.css" type="text/css" rel="stylesheet"> </head> <body> <h1>my Wepage</h1> <p>this is my webpage.</p> </body> </html> body { } h1 { } p { } background-color: #EBEBEB; font-weight: bold; font-size: 24pt; font-style: italic; 12

13 Creating PDFs Just like you saw in Data Camp Things that are different from what we normally do: All files should be in a markdown directory together Don't run rstudioapi code 13

14 Easiest Way to Create a Shiny App The simplest way to write a shiny app using Markdown: 1. Use write_csv() on your final ready-for-figure dataset in markdown folder 2. Have your original figure generation code ready 3. Create a new shiny markdown document using R Studio (not a new shiny app) 4. Delete everything below the YAML (unless you want an example) 5. Within a code block (```) add shiny functions for anything to be changeable Examples: numericinput(), radiobuttons(), selectinput(), sliderinput() 6. Test your shiny input code by copying each function's results into a.html file and viewing that file in a web browser to verify it looks right 7. Use read_csv() to import your final ready-for-analysis dataset 8. Copy existing figure generation code and modify to use shiny variables (remember to use input$ to refer to those variables) 9. Surround entire figure generation code with renderplot({ }) 14

15 ggplot2 vs ggvis A summary is available here: ggvis is a more advanced tidyverse plotting system that differs from ggplot2 in that ggvis: Is more experimental Combines elements using %>% instead of + Does not support faceting Distinguishes between static geoms (which are now emit_) and dynamic geoms (layers) Creates an HTML object when exported (vs. ggplot2's static image file) Enables animation using shiny and unique ggvis functions Is much faster Has less graphical complexity by default Has fewer emits than ggplot2 has geoms You generally want to use ggvis to render animations on webpages and either of the two (but probably ggplot2) for everything else 15

16 Easiest Way to Put a Shiny App Online Easiest: First time setup only Sign up for a free account (something.shinyapps.io) Follow the instructions on the webpage, shorthanded here 1. install.packages("rsconnect") 2. library(rsconnect) 3. On the website, click "Show Secret" then "Copy to Clipboard" paste that into R and run it 4. deployapp("nameofappfile.rmd") After initial setup library(rsconnect) deployapp("nameofappfile.rmd") 16

17 Common Problems with Shiny and shinyapps.io Forgot to set up access tokens in R Studio Forgot runtime:shiny Didn't put skinny datafile in markdown directory Didn't define shiny input functions correctly, especially related to named vectors Put too much re-processing within renderplot() 17

Web Scraping and APIs

Web Scraping and APIs http://datascience.tntlab.org Module 11 Today s Agenda A deeper, hands-on look at APIs A sneak-peak at server-side API code How to write API queries How to use R libraries to write